EOCSI EASI training session 3: Working with SAR data¶
This notebook will demonstrate how to use Sentinel-1 backscatter to detect the presence of water.
Introduction to Sentinel-1¶
Sentinel-1 carries a synthetic aperture radar (SAR) sensor, which can observe our planet even through cloud cover. Sentinel-1's SAR sensor is an active sensor; it emits light in the microwave range of the electormagnetic spectrum, and measures the amount that returns, known as backscatter. Smooth surfaces, like calm water, have very low backscatter - most of the light bounces away from the sensor (known as specular reflection). For rough surfaces, like dirt or vegetation, light is scattered in all directions, producing small backscatter signals (known as diffuse backscatter). Large, human-made structures, which feature both vertical and horizonal smooth surfaces, produce large backscatter signals (known as double bounce). As such, the intensity of Sentinel-1 backscatter can distinguish water from land, as shown in the image below:
In the image, the river appears dark (specular reflection), with the urban area appearing very bright (double bounce). For more information, see the article from GIS Geography on SAR.
This workshop¶
As of this workshop, Sentinel-1 backscatter is not yet available over South East Asia. Instead, this notebook will demonstrate how to load Sentinel-1 backscatter data over Africa, using Digital Earth Africa's STAC metadata. The Open Data Cube provides odc-stac, a useful library for loading data from STAC into an xarray, helping you work with it in the EASI platform. We will use this library in the workshop.
Important: If you want to work with Sentinel-1 as part of your project, please let us know and we can explore options to make data available to you.
The notebook contains the following steps:
- Import required packages for loading data and analysis
- Access Sentinel-1 analysis ready data from Digital Earth Africa
- Visualise data and export a multi-band, single time-slice GeoTIFF
- Applying speckle filter and converting the digital numbers to decibels (dB) values for analysis
- Use histogram analysis to determine the threshold for water classification
- Design a classifier to distinguish land and water
- Apply the classifier to the area of interest and interpret the results
